Entity Relationship Data Model - ecs.wgtn.ac.nz · Entity-Relationship Model - Terminology the Entity-Relationship model (ER, for short) is the most popular conceptual data model,

Lecturer: Dr Hui Ma

Engineering and Computer Science

Entity–Relationship Data Model

SWEN 304

Trimester 1, 2020

Database Design - Motivation

• Why do IT projects fail to meet customer expectations?

• There can be many reasons:

• the client is not sure what they want

• the requirements were not properly documented

• the lack of appropriate development methodology

• the desired functionality was difficult to develop

• the budget and resources were not sufficient

• the lack of team and work management

• . . .

SWEN304 Lect6and7: EERM 2

Database Design - Motivation


Database Design – Four Phases

• Database Design is the process of constructing a detailed schema of a databasethat can support the organization’s business needs and objectives for the databasesystem under development

• The database design process has four phases:

1 Requirements Collection and Analysis

2 Conceptual Design

3 Logical Design

4 Physical Design


Phase 1: Requirements Collection and Analysis

• Requirements collection and analysis is the process of collecting and analyzingdata requirements of the organization so as to provide database solutions thatfulfill business needs of the organization

• Compilation of data requirements includes:

• a description of the data to be used or generated

• details of how data is to be used or generated

• any additional requirements for database system under development

• . . .


Phase 2: Conceptual Design

• Conceptual design is the process of constructing a database schema used in anorganization, independent of any physical or implementation considerations

• modelling at a high-level of abstraction

• simple enough

• often with graphical representation

• used to communicate the logical structure of a database with domain experts andpotential users

• The data model is built using the input from the requirements specification

• Note: The conceptual design is based on the entity-relationship model in thiscourse


Phase 3: Logical Design

• Logical design is the process of constructing a database schema used in anorganization, as represented by a particular data manipulation technology, but stillindependent of physical considerations

• The conceptual database schema is mapped onto a logical database schema, whichcan be further refined to meet the data requirements

• popular data manipulation technogies used in logical design are those provided bythe relational model or some other model (for example: OO, XML, JSON)

• in case of the relational model, the database schema includes a description of alltables with their primary and foreign keys, so that SQL can be applied for datadefinition, data manipulation, and data querying

• Note: The logical design is based on the relational model in this course


Phase 4: Physical Design

• Physical design is the process of implementing the logical data model in adatabase management system (DBMS)

• If the logical data model is the relational model, then the physical design is toestablish relations in the DBMS that involves:

• selecting the files in which to store the tables

• deciding which indexes should be used to achieve efficient access to data items

• describing the integrity constraints and security measures

• . . .

• Decisions made during the physical design phase often affect non-functionalproperties of the database at run time, such as performance, accessibility, securityand user-friendliness

• Note: Details of physical design are beyond the scope of this course


Database Design Process - Overview


Entity-Relationship Model - Terminology

• the Entity-Relationship model (ER, for short) is the most popular conceptualdata model, and the de facto standard in conceptual design

• Originally proposed by Peter P. Chen in 1976

• The target of the database is regarded as consisting of entities and relationships

• Entities are basic objects that can be identified in the target of the database

• Attributes are properties that describe entities

• Entities that are described by the same set of attributes are classified into anentity set

• An entity set can be abstracted into an entity type

• Example:

A entity type Client with attribute set{Name, Birthday, Address, Phone}, andkey {Name, Birthday}

Client

–Phone–Address

–Birthday–Name


Entity-Relationship Model - Terminology

• Relationships are associations between entities, or objects that are derived fromentities

• Relationships have entities as their components

• Relationships can also have attributes as properties that describe them

• Relationships that are described by the same set of components and attributes areclassified into a relationship set

• A relationship set can be abstracted into a relationships type

• Example:

A relationship typeRental with componentset {Client, DVD},attribute set{RentalDay, DueDay},and key{DVD,RentalDay}

Client

–Phone–Address–Birthday–Name

DVDYear–

Director–Title–

RentalPP

PP��

�� —RentalDayPPPP—DueDayq

��

@@@R


Entity-Relationship Diagram

• For the communication between database designers, database programmers,domain experts and potential users, it is helpful to visualise the database schemagraphically

• A entity-relationship (ER) diagram is a directed graph with a node for everyentity type or relationship type

• Convention (proposed by Chen): draw entity types as rectangles, and drawrelationship types as diamonds

• To capture more details of the conceptual design, database designers often show ...

• attributes in the ER diagram by attaching them to their entity or relationship type

• keys in the ER diagram by underlining key attributes and by putting dots on keycomponents

• attribute domains in the ER diagram by adding them to the attributes


Entity-Relationship Diagram - Example

• An example:


Entity-Relationship Schema - Example

• In our example we have the following entity types:

• Supplier

• attributes: supplier no, name, address

• key: supplier no

• Article

• attributes: article no, price, nr on stock, short name

• key: article no

• Bill

• attributes: bill no, amount

• key: bill no

• Department

• attributes: department no, budget

• key: department no

• Client

• attributes: client id, name, address

• key: client id


Entity-Relationship Schema - Example

• In our example we have the following relationship types:

• Offer

• components: Supplier, Article

• attributes: date, price

• key: Supplier, Article, date

• Purchase

• components: Supplier, Article, Department

• attributes: date, quantity

• key: Supplier, Article, Department, date

• Sale

• components: Department, Article, Client, Bill

• attributes: date, price

• key: Department, Article, Client, date


Entity-Relationship Schema

• An entity-relationship schema (ER schema, for short) is a finite set S of entityand relationship types, such that for every relationship type R in S all itscomponents belong to S, too

• An instance St of an ER schema S assigns each entity type E an entity set Et

and each relationship type R a relationship set Rt, such that for each relationshipr in the relationsip set Rt and for every component E of R the entity r(E)belongs to the entity set Et


Keys

• Keys ensure that the entities in an entity set (or the relationships in a relationshipset) can be uniquely identified

• In our example the entity types have keys as follows:

• clients are uniquely identified by their client id

• departments are uniquely identified by their department no

• bills are uniquely identified by their bill no

• articles are uniquely identified by their article no

• suppliers are uniquely identified by their supplier no

• In our example, the relationship types have keys as follows:

• offers are uniquely identified by supplier, article and date

• purchases are uniquely identified by department, supplier, article and date

• sales are uniquely identified by article, department, client and date


n-ary Relationship Types

• An entity type E is described by its attributes attr(E) and its key key(E)

• A relationship R is described by its components comp(R), its attributes attr(R)and its key key(R)

• A relationship type with n components is called n-ary, in particular

• a unary relationship type has one component

• a binary relationship type has two components

• a ternary relationship type has three components


Recursive Relationship Types

• A relationship type is called recursive if it contains the same entity type more thanonce as a component

• Example: suppose we want to model relationships between a child and a parent(that is, between two persons)

• Can use a relationship type Descendent that contains the entity type Persontwice as a component

Person–Address

–Birthday–Name

Descendent

PPPPP

��

��PPPPP

Child Parentr r? ?

• Role names are used to distinguish the different occurrences of the same entitytype in a recursive relationship type (in our example Child and Parent)


Extentions of the Entity-Relationship Model

• Several extensions have been proposed to overcome inconveniencies and to enrichthe expressive power of the ER model

• After all, the ER model with its extensions provides effective and convenient meansof describing the target of the database at a high level of abstraction

• Some extensions we discuss here:

• Specialisation

• Higher-order relationship types

• Generalisation and clusters


Specialisation

• Sometimes, an object in the target of the database can be represented by morethan just a single abstract concept

• Examples:

• Students are also persons

• Graduate students are also students, and thus hey are also persons

• Abstract concepts can be derived from other abstract concepts, for example themore specific concept Student from the more general concept Person

• This idea is known as specialisation:

• the derived object type is a subtype of the more general supertype

• the subtype inherits all features of the supertype, but often adds some additionalproperties

• the subtype may be modelled as a unary relationship type whose single componentis just the supertype (but may have additional attributes)

• Note: every object of the subtype gives rise to at most one object of supertype


An Example for a Specialisation Hierachy


Same Example with further Relationship Types


An Example for Aggregation

• The following ER diagram contains two relationship types Patient and Driver

• Suppose we want to model that a person is in hospital because she/he had anaccident with a vehicle


An Example for Aggregation

• Can use a new relationship type Accident to model these relationships

• But what are the components of Accident?

• It would be convenient to regard parts of the ER diagram as entity types (thisapproach is called aggregation)


An Example of Relationship Types participating in Relationship Types

• In fact, we could simply use the relationships types Patient and Driver ascomponents of Accident

• It would be convenient to allow relationship types to participate in otherrelationship types (these are called higher-order relationship types)

• In our example the relationship types have the following order:

• Patient and Driver have order 1

• Accident has order 2


Higher-order Relationship Types

• A relationship type R is of order 1 if all its components are entity types

• Recall: In a (simple) ER diagram all relationship types are of order 1

• A relationship type R is of order k if its components have maximum order k − 1

• This extension is possible in an extended ER diagram

• It gives the database designer more flexibility, for example to model aggregation

• For the sake of convenience, entity types and relationship types together all calledobject types

• Entities and relationships are called objects

• Entity sets and relationship sets are called object sets

• Entity types may be regarded as object types of order 0

• Object types of order k are just relationship types of order k


Cluster

• sometimes it is necessary to model alternatives, for example having a relationshipto students or lecturers

• Abstract concepts can be obtained by generalization from a collection of abstractconcepts

• A cluster C represents an alternative among a collection of object types

• let C be an alternative between given object types C1, . . . , Cm

• for simplicity, the cluster can be denoted as C = C1 ⊕ · · · ⊕ Cm

• we call C1, . . . , Cm the components of the cluster C

• Example: Employee = Lecturer⊕Tutor⊕General Staff

• Every object of cluster type C is an object of exactly one of the object types in thealternative

• A cluster is of order k if its components have maximum order k − 1


An Example for a Cluster


An Example for a Cluster participating in a Relationship Type


Extended Entity-Relationship Schema and Diagram

• An extended entity-relationship schema (extended ER schema, for short) is afinite set S of object types (entity types, relationship types, clusters), such that forevery object type O in S all its components belong to S, too

• An instance St of an extended ER schema S assigns each object type O an objectset Ot, such that for each object o in the object set Ot and for every componentO′ of O the object o(O′) belongs to the object set O′t

• A extended entity-relationship diagram (extended ER diagram, for short) is adirected graph with a node for every entity type or relationship type

• Convention: draw entity types as rectangles, draw relationship types as diamonds,draw clusters as ⊕


Critical Success Factors in Database Design

• Work interactively with domain experts and users as much as possible

• Follow a structured methodology throughout the data modelling process

• Employ a data-driven approach

• Incorporate structural and integrity considerations into the data models

• Combine conceptualisation, refinement and validation techniques into the datamodelling methodology

• Use diagrams to represent as much of the database schema as possible

• Build a data dictionary to supplement the diagrams

• Be willing to repeat steps


Transformation EER → Relational (1)

• we describe the transformation of an EER schema into a relational databaseschema

• start with entity types (object types of order 0) and then work one’s way upgradually

• Transformation of entity types:

• an entity type E = (attr(E), key(E)) leads to a relation schema E′

• the attributes of E are preserved

• the domain assignment for the attributes of E is preserved, too

• the key of E becomes the primary key of the relation schema E′


Example - Transforming Entity Types

• three entity types:

• Department = ({No,Budget}, {No})• Supplier = ({No,Name,Address}, {No})• Article = ({No,Shortname,QuantityOnStock}, {No})

• result in three relation schemata:

• Department = {No,Budget} with primary key {No}• Supplier = {No,Name,Address} with primary key {No}• Article = {No,Shortname,QuantityOnStock} with primary key {No}


Transformation EER → RDM (2)

• for a relationship type R = (comp(R), attr(R), id(R)) and each componentC ∈ comp(R) choose pairwise disjoint sets of new attribute names not occurringin attr(R):

k attr(C) = {C.A where A is a key attribute of C′}

where C′ is originating from a prior transformation of component C

• Example: we derive new sets of attribute names for entity types:

• k attr(Department) = {Department No}• k attr(Supplier) = {Supplier No}• k attr(Article) = {Article No}



• Transformation of relationship types:

• a relationship type R = (comp(R), attr(R), id(R)) leads to relation schema R′ with

attr(R′) =⋃

C∈comp(R)

k attr(C) ∪ attr(R)

• the domain assignment of the attributes is preserved

• R′ has the primary key ⋃C∈key(R)∩comp(R)

k attr(C) ∪ (key(R) ∩ attr(R))

• each component C ∈ comp(R) defines a foreign key

[C.A1, . . . , C.An] ⊆ C′[A1, . . . , An]

on R′ for key(C) = {A1, . . . , An}


Example - Transforming Relationship Types

• Recall:

• k attr(Department) = {Department No}• k attr(Supplier) = {Supplier No}• k attr(Article) = {Article No}

• relationship typeDayOffer = ({Supplier,Article}, {Date,Price}, {Supplier,Article,Date})becomes:

• DayOffer = {Supplier No,Article No,Date,Price} with

• primary key: {Supplier No,Article No,Date}• foreign keys: [Supplier No] ⊆ Supplier[No]

[Article No] ⊆ Article[No]


Example - Transforming Relationship Types

• first we derive new sets of attribute names for components of relationship typePurchase:

• k attr(DayOffer) ={DayOffer Supplier No,DayOffer Article No,DayOffer Date}

• then relationship type Purchase =({Department,Day Offer}, {Quantity}, {Department,Day Offer})

becomes:

• Purchase = {Department No,DayOffer Supplier No,DayOffer Article No,DayOffer Date,Quantity} with

• primary key:{Department No,DayOffer Supplier No, DayOffer Article No,DayOffer Date}

• foreign keys: [Department No] ⊆ Department[No][DayOffer Supplier No,DayOffer Article No,DayOffer Date]⊆ DayOffer[Supplier No,Article No,Date]


How to handle Clusters

• cluster types used in conceptual design to model alternatives

• the relational data model does not provide similar concept

• transform EER schema with clusters into equivalent EER schema without clusters

• only necessary as pre-processing before actual transformation to the realationaldata model

• in general: clusters provide database designers a convenient way to model objectsin the target of database

• do not recommend to avoid clusters as size of EER schema increases dramaticallyand becomes harder to comprehend



• replacing clusters in EER schema is straightforward:

• cluster types in EER schema S that are not component of any relationship type canbe removed from S

• consider relationship type R with cluster component C = C1 ⊕ · · · ⊕ Cn

• replace R by n new relationship types R1, . . . , Rn:

• for i = 1, . . . , n: Ri obtained from R by replacing every occurrence of C by Ci

• if Ri still contain clusters, then repeat process of replacing these clusters by itscomponents

• the final EER schema is cluster-free and previous transformation to RDM can beapplied


Example - Transforming Clusters

• consider the following EER diagram:

Associate Prof

Professor

GraduateUnder

Project

Student

since until

PostGrad

DegreeSupervise

SupervisorDept Subject

Dept Level

Name

Name Title Description

ID Name

• Supervise=({Supervisor, Student, Project},{since, until},{Supervisor,Project}) with

• cluster Supervisor = Professor⊕AssociateProf

• cluster Student = PostGrad⊕UnderGraduate


Example - Transforming Clusters

• using Supervisor we obtain:

• Prof Supervision:({Professor, Student, Project},{since,until},{Professor, Project})

• AProf Supervision:({AssociateProf, Student, Project},{since,until},{AssociateProf,

Project})

• using Student subsequently we obtain:

• Prof PostGrad Supervision:({Professor, PostGrad, Project},{since,until},{Professor, Project})

• Prof UnderGrad Supervision:({Professor, UnderGraduate, Project},{since,until},{Professor,

Project})• AProf PostGrad Supervision:

({AssociateProf, PostGrad, Project},{since,until},{AssociateProf,Project})

• AProf UnderGrad Supervision:({AssociateProf, UnderGrad, Project},{since,until},{AssociateProf,

Project})


Entity Relationship Data Model - ecs.wgtn.ac.nz · Entity-Relationship Model - Terminology the Entity-Relationship model (ER, for short) is the most popular conceptual data model,

Documents